Subsampling design

SubsamplingDesign

Returned by: JinkoClient.get_subsampling_design, JinkoClient.iter_subsampling_designs, JinkoClient.list_subsampling_designs, SubsamplingDesign.edit, SubsamplingDesign.set_trial, Trial.create_subsampling_design

Also has every member of ProjectItem.

Member	Kind	Description
`generated_vpops`	attribute
`numeric_filters`	property
`categorical_filters`	property
`marginals`	property
`categoricals`	property
`correlations`	property
`summary_statistics`	property
`survivals`	property
`observables`	property
`source_trial`	property	Return the trial snapshot currently linked to this subsampling design.
`diagnostics`	property	Return sanity diagnostics for the current subsampling design snapshot.
`diagnostics_at`	method	Return sanity diagnostics for a specific subsampling design revision.
`estimate_distributions`	method	Estimate candidate target distributions for a source-Trial scalar.
`generate_vpop`	method	Generate a Vpop from this subsampling design.
`edit`	method	Advanced partial update for this subsampling design.
`set_trial`	method	Replace the source trial for this subsampling design.

generated_vpops

source_trial

Type: Trial

Return the trial snapshot currently linked to this subsampling design.

diagnostics

Type: SubsamplingDesignDiagnostics

Return sanity diagnostics for the current subsampling design snapshot.

Each access fetches a fresh, self-consistent snapshot: sanity messages plus the resolved trial descriptor context used by SubsamplingDesignDiagnostics.explain. Filtering the returned view (.errors(), .for_field(...), etc.) and calling .explain() do not make further requests - re-access this property for updated results.

diagnostics_at

diagnostics_at(revision: int) -> SubsamplingDesignDiagnostics

Return sanity diagnostics for a specific subsampling design revision.

See diagnostics for the freshness contract of the returned view.

estimate_distributions

estimate_distributions(
    scalar_id: str, *, arm: str | None = None
) -> openapi_types.SubsamplingEstimateResponse

Estimate candidate target distributions for a source-Trial scalar.

Parameters:

Name	Type	Description	Default
`scalar_id`	`str`	Identifier of a scalar exposed by the source Trial.
`arm`	`str \| None`	Optional source-Trial arm for arm-specific scalar outputs.	`None`

Returns:

openapi_types.SubsamplingEstimateResponse: Candidate fitted laws and compatible subsampling target forms. The
openapi_types.SubsamplingEstimateResponse: result is informational; choose a target only after scientific
openapi_types.SubsamplingEstimateResponse: review.

generate_vpop

generate_vpop(
    *,
    boltzmann_constant: float,
    iters_fixed_temperature: int,
    num_iterations: int,
    num_samples: int,
    replacement_rate: float,
    seed: int,
    folder: Folder | str | None | _UnsetType = _UNSET,
    name: str | None = None,
    description: str | None = None,
    version: str | dict | None = None
) -> Vpop

Generate a Vpop from this subsampling design.

Parameters:

Name	Type	Description	Default
`boltzmann_constant`	`float`	Controls acceptance probability for uphill moves in the simulated-annealing algorithm. The API default is `1e-3`.
`iters_fixed_temperature`	`int`	Number of iterations at each temperature level.
`num_iterations`	`int`	Total number of simulated-annealing iterations. Must be >= `iters_fixed_temperature`.
`num_samples`	`int`	Number of patients to subsample from the source trial.
`replacement_rate`	`float`	Proportion of samples swapped at each iteration (between 0 and 1).
`seed`	`int`	Integer seed for the random-number generator, ensuring reproducibility.
`folder`	`Folder \| str \| None \| _UnsetType`	Destination folder for the generated Vpop. Defaults to the same folder as this subsampling design.	`_UNSET`
`name`	`str \| None`	Optional display name for the generated Vpop.	`None`
`description`	`str \| None`	Optional description.	`None`
`version`	`str \| dict \| None`	Optional version label.	`None`

edit

edit(
    *,
    trial: Trial | _UnsetType = _UNSET,
    numeric_filters: Sequence[dict[str, Any]] | _UnsetType = _UNSET,
    categorical_filters: Sequence[dict[str, Any]] | _UnsetType = _UNSET,
    target_marginals: Sequence[dict[str, Any]] | _UnsetType = _UNSET,
    target_categoricals: Sequence[dict[str, Any]] | _UnsetType = _UNSET,
    target_correlations: Sequence[dict[str, Any]] | _UnsetType = _UNSET,
    target_survivals: Sequence[dict[str, Any]] | _UnsetType = _UNSET,
    target_summary_statistics: Sequence[dict[str, Any]] | _UnsetType = _UNSET,
    additional_scalars: Sequence[str | dict[str, Any]] | _UnsetType = _UNSET,
    version: str | dict | None = None
) -> SubsamplingDesign

Advanced partial update for this subsampling design.

Omitted fields keep their current values. This method keeps the backend field names for compatibility with lower-level callers. For day-to-day editing, prefer the typed subservices such as design.marginals or convenience wrappers like set_trial(...).

Parameters:

Name	Type	Description	Default
`trial`	`Trial` `\| _UnsetType`	Replacement source trial.	`_UNSET`
`numeric_filters`	`Sequence[dict[str, Any]] \| _UnsetType`	Full replacement for `numericFilters`.	`_UNSET`
`categorical_filters`	`Sequence[dict[str, Any]] \| _UnsetType`	Full replacement for `categoricalFilters`.	`_UNSET`
`target_marginals`	`Sequence[dict[str, Any]] \| _UnsetType`	Full replacement for `targetMarginals`.	`_UNSET`
`target_categoricals`	`Sequence[dict[str, Any]] \| _UnsetType`	Full replacement for `targetCategoricals`.	`_UNSET`
`target_correlations`	`Sequence[dict[str, Any]] \| _UnsetType`	Full replacement for `targetCorrelations`.	`_UNSET`
`target_survivals`	`Sequence[dict[str, Any]] \| _UnsetType`	Full replacement for `targetSurvivals`.	`_UNSET`
`target_summary_statistics`	`Sequence[dict[str, Any]] \| _UnsetType`	Full replacement for `targetSummaryStatistics`.	`_UNSET`
`additional_scalars`	`Sequence[str \| dict[str, Any]] \| _UnsetType`	Full replacement for `additionalScalars`.	`_UNSET`
`version`	`str \| dict \| None`	Optional version label for the new snapshot.	`None`

set_trial

set_trial(
    trial: Trial, *, version: str | dict | None = None
) -> SubsamplingDesign

Replace the source trial for this subsampling design.

All other fields keep their current values. See edit for the full parameter reference.